EPA Methods 306, 306A, and 306B have recently been revised to incorporate the new EPA/EMMC¹ format that is required in codifying future analytical monitoring methods. While revising EPA Method 306 entitled, "Determination of Chromium Emissions from Decorative and Hard Chromium Electro-plating and Chromium Anodizing Operations - Isokinetic Method" sampling and analytical guidance was added to clarify and update the technical procedures.

California Air Resources Board (CARB) Method 425 entitled "Determination of Total Chromium in Hexavalent Chromium Emissions from Stationery Sources" that measures hexavalent and/or total chromium was also modified recently by CARB.

The revised methods have been peer-reviewed by EPA analytical staff experienced in inorganic chemistry and/or related procedures associated with the measurement of total and/or hexavalent chromium (Cr+6). The primary objectives of the peer review were to (1) perform a quality assurance/ quality control (QA/QC) review of EPA Method 306 and CARB Method 425 analytical differences, defects, and similarities; (2) summarize the findings of the review; and (3) provide detailed recommendations for corrections, modifications, and improvements.

This report provides a detailed summary of the evaluation as well as an overview of the measurement techniques and specific differences that exist between the EPA 306 and CARB 425 methods. Comments and recommendations regarding the individual test methods are also summarized.

EPA Method 306 provides three (3) analytical techniques for the measurement of the wide-range of chromium emissions found at hard chromium electro-plating and anodizing sources. These techniques are inductively coupled plasma emission spectrometry (ICP), graphite furnace atomic absorption spectrometry (GFAAS), and ion chromatography with post column reaction (IC/PCR). The ICP and GFAAS analytical methods are applicable for the measurement of total chromium only. The IC/PCR analytical method is applicable for the measurement of speciated (trivalent and hexavalent) chromium emissions.

CARB Method 425, allows the analyst to use either the GFAAS or IC/PCR technique but does not provide for the use of the ICP technique. Instead, the analyst has, as an option, the use of a manual colorimetric technique that uses the same color reaction as the IC/PCR determination, but with significantly less sensitivity. However, EPA only allows the use of the manual colorimetric approach with certain limitations.

Since it is expected that chromium emissions from the chrome plating sources will be almost exclusively hexavalent chromium (Cr+6), the first two analytical methods are referenced for the more routine analytical measurements. In the event that chromium speciation is required, IC/PCR is provided for determination of hexavalent chromium. Relative strengths and weaknesses for the three referenced techniques are discussed in the following sections.

Inductively coupled plasma emission spectrometry is a single or multi-element emission technique that has become routine over the last fifteen years. The method has a wide range of linearity (up to 5 orders of magnitude) and is reasonably sensitive; a chromium detection limit of <5 µg Cr/L to 30 µg Cr/L can usually be obtained, depending on the emission source. EPA has designated that this technique be used to measure only high chromium concentrations >35 µg Cr/L when demonstrating compliance . The ICP determination is probably the most rapid of the three techniques, allowing the analyst to perform a sample measurement in about two minutes. The method is especially rapid since no sample preparation (digestion) is required.

The most recent improvements include the use of an "axial" torch which provides approximately a tenfold improvement in sensitivity over the more traditional radial torch design. With the axial design, the analytical signal for the chromium is enhanced, however, the analytical signals for the interfering species are similarly enhanced. For the difficult sample matrix (high dissolved solids) associated with the Method 306 alkaline medium, this may present as many difficulties as advantages. ICP techniques possess a number of interferences that are worthy of note:

Spectral interferences due to iron, manganese, and uranium are possible. Generally speaking, these interferences can be corrected through the use of an alternate wavelength or by automated interfering element corrections provided by the instrument software. If it can be

verified that these species do not exist at significant concentrations in the sample, it is suggested that these correction techniques not be used since they tend to result in some degradation of the detection limit.

Background interference due to a matrix mismatch between samples and standards may be encountered and corrected. However, the background interference correction may not be needed since the method requires that a alkaline sampling medium be used in the calibration standard preparation. For these alkaline samples, perhaps the most significant interferences are physical, which can be attributed to a high dissolved solids content in the sample. There tends to be an accumulation of solid material from the alkaline salt at the end of the plasma torch (axial or radial) which severely compromises the analytical results, as well as greatly decreases the lifetime of the torch. Sophisticated ICP sample introduction techniques such as the ultrasonic nebulizer are not recommended since they concentrate the high solids matrix in addition to concentrating the Cr analyte prior to injection into the plasma.

This technique allows the analyst to accurately determine Cr concentrations in the impinger samples that are <35 µg Cr/mL. This technique has been used reliably for more than 20 years by analytical laboratories. Chromium concentrations that are <1 ppb(µg/L) can be detected with this technique. Quality measurements are, however, highly dependent on the skill of the analyst, although less so with the improved furnace designs and computer controls provided with instrumentation during the last 15 years. The GFAAS technique is not ideally suited to the 306 and 425 Methods’ alkaline samples, but with reasonable care, quality data can be obtained. The GFAAS technique is not as rapid as the ICP procedure; with analysis times ranging up to four minutes.

The sample is digested in nitric acid (HNO₃) prior to measurement and sample intensity is compared to that of calibration standards prepared in HNO₃. This is the only procedure of the three techniques that requires a significant sample preparation operation. In normal laboratory operation, the calibration standards do not contain the alkaline reagent. Therefore, the signal of the sample may be suppressed with respect to that of the less complex standard matrix, thereby requiring the need for some type of matrix compensation technique. This may be accomplished by (1) adjusting the furnace conditions; (2) adding alkaline reagent to the standards to match the matrix of the sample; or (3) resorting to the method of standard additions (MSA). Frequently the first approach is all that is required. Also graphite furnace tubes differ considerably in quality which tends to affect both the instrument performance as well as the analytical strategy employed with the matrix correction issues.

GFAAS is particularly prone to spectral interferences due to smoke resulting from the dissolved solids in the sample. Due to the severity of the sample matrix, a background correction system may be necessary, such as the Zeeman correction system. Many instruments manufactured within the last fifteen years apply this technique automatically for sample measurements. Another acceptable system uses the Smith-Hieftje correction. Methods that employed the Deuterium background correction are generally not considered effective in this analytical application.

Dedicated GFAAS instrumentation produced in the last 15 years are normally equipped with autosampling features. Sample volumes ranging from <5µL to 50µL and sometimes up to 100µL are pipetted accurately and with excellent precision. This repetition is essential to quality GFAAS measurements. Early techniques requiring manual sample pipetting into the furnace were greatly dependent on the technique of the analyst. Consequently, duplicate injections were required for reliable results. This is no longer the case with current autosampling systems.

The alkaline medium encountered during these measurements produces considerable degradation of the graphite parts resulting in much shorter graphite tube and contact ring lifetime; 10%-30% of normal duration. It is for this reason that matrix matching of the standards with samples should be a last resort since this will result in additional analyses of solutions containing highly dissolved solids.

This technique is probably the most sensitive of the three instrumental methods and is certainly the most specific with respect to hexavalent chromium measurements. Current instrumentation allows the chemist to detect Cr⁺⁶ levels of <0.05 µg/L in the alkaline solution. Early approaches employed the use of a pre-concentrator column packed with the same ion exchange resin used in the analytical column. With improved detector design and superior materials of construction (Teflon instead of stainless steel), detection limits have dropped significantly and the need for the pre-concentration approach is only required with older instrumentation. While no significant sample preparation is required for this technique, filtration of the sample may be necessary to prevent clogging of the sample injection system. The actual measurement time for each sample is probably the longest for the three methods - greater than 5 minutes per sample.

IC/PCR is an extremely reliable technique for the determination of the hexavalent Cr species in the presence of the trivalent form of Cr (Cr⁺³). The technique has, for all practical purposes, no significant interferences. The hexavalent chromium ion is isolated on the analytical column and further specificity is provided through the reaction of the Cr⁺⁶ ion with the color agent, diphenycarbazide. The colorimetric measurement is made at the 540 nm wavelength. The likelihood of a species both co-eluting with the Cr⁺⁶ ion and reacting to form a sensitive color complex at the 540 nm wavelength is remote.

This method is perhaps more robust than ICP or GFAAS in handling the alkaline medium and does not experience the hardware deterioration to which the previously described methods are prone.

Early systems did require some skills in assembly of the analytical system. "Homemade" systems can now be constructed from HPLC pumps, a sampling valve, and an analytical column, with the only major purchase being the spectrophotometer (colorimeter) detector. Such systems can cost <$10,000. Complete systems, however, are now available commercially that include an autosampler, the analytical system and a data acquisition system.

Quality control activities differ slightly between the measurement methods, not so much due to the instrumentation but rather due to the sample preparation techniques that precede the measurement. Methods 306 and 425 address, to varying degrees, accuracy and precision issues for all the measurement techniques. Quality control requirements for each method and measurement technique are indicated in Table 1. Included in Table 1 are QA/QC criteria for the original Method 306 and for the revised Method 306.

Method accuracy is assessed through the use of spikes and standard reference materials (SRMs). Analytical accuracy is assessed for both the measurement phase and for the overall analytical method. In this way, the analyst can pinpoint the source of any potential errors. Accuracy is expressed as percent of recovery based on "True" or expected value. In the case of the ICP or IC/PCR techniques, measurement recovery and overall method recovery are essentially one and the same, since there is no sample digestion procedure employed. For GFAAS there is a distinct sample digestion step involved and recovery values will reflect errors in both the digestion and the analytical measurement steps. In addition, it is critical that the accuracy of the calibration standards be established immediately following the calibration procedure (prior to measurement of field samples). The analytical measurement accuracy is assessed through the use of spikes that are added to the sample immediately prior to its measurement. Poor recoveries indicate that species/substances within the sample matrix are interfering with the excitation process of the chromium species in the plasma. Common ways to compensate for this effect is through one or more of the following approaches:

1) optimization of instrument conditions

2) matrix-matching of standards with samples (not always possible)

3) implementation of a Method of Standard Additions (MSA) technique

4) addition of an internal standard to samples and standards

The latter approach is not particularly feasible with the GFAAS and IC/PCR techniques but could be used with the ICP determination. This approach requires extensive knowledge of the spectroscopy operation and should not be considered routine laboratory operation. Method 306 notes the use of the first three approaches for ICP measurements. Standards (calibration and QC) should be prepared in the alkaline reagent of choice and matrix interferences should therefore be minimal.

Methods 306 and 425 address GFAAS calibration in essentially the same manner, that is, strict matrix matching is not necessarily required. Standards and digested samples both contain 1 percent HNO₃ but only the field samples contain the alkaline reagent. It is implied that careful adjustment of instrument conditions will compensate for matrix differences. While this is true in many cases, matrix compensation can be dependent on the quality of the particular lot of graphite parts in the furnace. Both methods refer to the option of the MSA approach for matrix compensation..

Some laboratories have reported discrepancies between Method 306 for total Cr and hexavalent Cr measurements for the same source sample. This is quite possible if matrix corrections for the total Cr determination by either ICP or GFAAS are not carefully performed. While the total Cr determination can be within Method 306 specifications, the total Cr result may still be biased low with respect to the Cr+6 result by as much as 15 percent since Cr⁺⁶ determinations using IC/PCR are not particularly prone to matrix effect/suppression. The same phenomenon can occur with GFAAS total Cr determinations.

The source solution used to prepare the calibration standards as well as the accuracy of the preparation procedure itself should be assessed through the use of calibration reference standards (also known as Initial Calibration Verification {ICV} standards in many reference methods). This solution must be prepared from an entirely different source (commercial or otherwise) and is analyzed immediately following the calibration. Usually, either 5 percent or sometimes 10 percent agreement with expected values are considered acceptable. This approach should be used for all analytical techniques.

On a related topic, accuracy can be compromised if there is instrumental drift over the course of the analysis. Drift may be assessed by analyzing a standard or sample of known concentration at specific intervals during a sample analysis run. Normally, the drift may be considered excessive if the measured value of the standard used differs from the expected value by more than 10 percent. Drift check standards are also known as Continuing Calibration Verification (CCV) standards.

Accuracy can be affected if there is a drift in the "zero" point of the calibration curve. This usually is caused by some contamination in the calibration blank and is assessed immediately following calibration (otherwise known as the Initial Calibration Blank or ICB) as well as throughout the analysis run (otherwise known as the Continuing Calibration Blank). In addition, proper adjustment of sample "washout" times between samples may be verified through the periodic analysis of the calibration blank.

As with accuracy, both measurement and overall method precision must be determined. Measurement precision is determined through the use of replicate injections of the sample solution. Overall analytical method precision is determined by processing two portions of the sample through the entire laboratory (preparation and analysis) procedure. It is expected that the overall method position will require a wider acceptance criteria since it involves several additional steps, each of which contribute a certain error to the determination.

The ICP and IC/PCR techniques require duplicate "injections" of the same sample since no sample preparation technique is required. Unlike the other methods, duplicate digestions of the same sample are performed when using the GFAAS determination. Both Method 306 and CARB 425 are consistent on these QC requirements.

Interference checks are used primarily with the ICP measurement. A variety of spectral interferences due to interfering elements are possible with this technique. This phenomenon may be assessed through the analysis of known standards containing the expected interfering element(s) and measuring the "false" analyte (Cr) signal attributable to the interfering element. The interference may be corrected by implementing the instrument software.

Interferences exist for the other techniques also. Normally, spectral interferences are encountered with GFAAS but are usually corrected with the instrumental background correction systems.

With IC/PCR, a co-eluting species could theoretically complex with the diphenylcarbazide reagent causing a chromatographic interference but this likelihood is small. Occasionally peaks may appear in the vicinity of the chromate peak that are of uncertain identity. The analyst must then re-inject the sample that has been spiked with additional Cr. If the combined peak is of a satisfactory shape (no doublets, etc.) then the original peak may be assumed to be Cr and the system may be undergoing drift which would require a subsequent corrective action. If, however, multiple peaks are observed, and clearly resolved, and the proper retention time is observed for the Cr spike, then the system is considered to be operating properly.

Analytical detection limits are ordinarily determined through the measurement of replicate aliquots of a sample or standard with an analyte concentration near the expected LOD. Replicate measurements of a laboratory or field blank are also accepted by some agencies. For samples containing a difficult matrix, it is important that a low concentrated solution contain a representative amount of the matrix solution. It is also critical that all applicable digestion/extraction steps be implemented in the preparation of each LOD replicate. For example, each replicate would require a separate digestion for the GFAA determination of total Cr.

While both methods approve the use of the GFAAS and IC/PCR techniques, they differ in the techniques used for measuring higher concentrations of Cr emissions (>35 µg Cr/L). Method 306 specifies the ICP technique, and assumes that the hexavalent Cr species are the predominant form of Cr being emitted from the chrome plating facilities.

Method 425 is designated for a less specific application, namely, any stationary source. Therefore, CARB approves the use of a classical manual colorimetric technique, which provides a reasonable measure of compound specificity. This latter technique, incidentally, is considerably more prone to interferences from other chemical species than the IC/PCR technique.

While the Methods are similar in their treatment of the measurement techniques, there are some differences in their approach to the quality control / quality assurance issues. These are outlined in Table 1. Method 306 provides significantly more QA/QC acceptance criteria for the analytical data than does Method 425. Although the GFAAS QC procedures are covered in some detail in Method 425, no QC acceptance criteria are provided. Greater detail is provided for the IC/PCR measurement, but still is far less than that described in Method 306.

Method 425 provides the analyst with a detailed detection limit calculation (3.4.1) that is consistent with the overall approach to the "Pre-Test Protocol" designed to ensure that a sufficient sample volume is collected. However, the calculations are based on what appears to be a faulty premise. The method describes a calculation based on the standard deviation of the mean of at least four replicates of a standard concentration near the mid-point of the curve (Section 3.4.1). The method should have specified a calibration curve constructed near the lower end of the calibration curve, not the curve used for actual field sample analysis. Alternatively, CARB could have specified the analysis of a low standard/sample (one that is approximately 5-10 times the expected LOD) and changed the term s_mid to read s_low. In a contradictory statement, however, Method 425 also mentions the use of at least four "lab blanks" for the determination of the LOD (Section 17.1.1). Method 306 quotes LODs from SW-846 measurement methods without specifying calculations/procedures. An attempt has been made to clarify this issue in the Method 306 revision.

In the revised version of Method 306, an attempt has been made to provide criteria for all of the QC operations. In addition, a brief reference to the detection limit calculation is provided.

This report attempts to summarize issues related to the techniques themselves, including advances in recent years and data quality issues that are specific to each particular measurement approach.